Skip to content

quest: plan asking the maintainer only what their answer changes - #696

Closed
kixelated wants to merge 3 commits into
mainfrom
quest-plan/ask-less
Closed

kixelated wants to merge 3 commits into
mainfrom
quest-plan/ask-less

Conversation

@kixelated

@kixelated kixelated commented Nov 13, 2026 •

Copy link
Copy Markdown
Collaborator

Plans the ask-less epic: agents ask the maintainer only the decisions their answer is likely to change. Everything else proceeds on a recorded standing rule and stays reviewable after the fact.

Evidence, from quest messages on the maintainer's three Linux hosts (2026-10-20 to 11-13), their session logs since 11-03, and the PR timelines of five repositories:

  • About 500 answers to blocking questions a day, 74% taken unchanged. When the agent recommended merging, 95% were.
  • Waiting from ready to the first merge pick is 40% of merged PRs' summed lead time (median 1.0 h, p90 12.2 h).
  • 425 answers repeated a pick or skip already given for the same PR.
  • The maintainer's overrides cluster in goal and scope, API shape and naming, and cutting or deleting work.

Changes

New epic quest/m0/ask-less/, holding the baseline and every decision:

  • [M] standing-picks: merge everything outside a protected list on a standing pick
  • [M] recorded-picks: record picks and skips on the PR; a skip holds until the head changes
  • [M] landed-list: list what landed on a standing pick, and act on it (requires recorded-picks)
  • [S] milestone-by-rule: place new quests by rule instead of asking
  • [L] headless-loop: run /quest-run headless and run a configured command when only decisions wait
  • [L] decisions-session: one /quest-complete across every repository
  • [S] messages-codex-tui: quest messages fails on Codex 0.162 TUI turns on two of three hosts

quest/m0/mistake-audit.md now also counts escapes by standing and maintainer pick, per class.

Host work went to kixel-ai/dotfiles' quest tree: the loop service with its Discord push, and a dev-shell guard fix.

Decisions

All recorded in the epic's and quests' Plans. One reversal: the dev-shell guard was first picked as "any merged flake version, whatever HEAD is". The maintainer then kept the HEAD-on-origin check once it turned out that dropping it reopens the fork-PR hole kixel-ai/dotfiles@7202fdf closed.

Follow-ups

  • Drop: making agents' questions explain their terms. 13% of typed overrides asked "what is" or "why", but the guide's Questions section already requires it, and the mistake audit tracks it.

Validation

After merging main, quest check, just check, and just test pass on the head. The merge dropped two links to quest/m0/agent-admission.md, which #692 finished; headless-loop now notes that quest agent start admits itself. No behavior to exercise: this PR changes only plans.

Public API

None.

Models

Size: M
Implementer: Claude Code, claude-opus-5-5, unknown effort

The research and interview ran on claude-fable-5-2 in the same session.

🤖 Generated with Claude Code

(Written by Claude Opus 5.5)

Review

Rewritten by quest review after each round.

  • Head: 8752084

  • Base: origin/main at cd70476

  • Reviewer: Codex, gpt-6.1-sol, high effort

  • Session: 01a1d6d5-0325-7bb1-a654-e0bd19facec1, round 1

  • Score: 10/10, The diff contains coherent planning documents and audit guidance only, with no runtime changes or actionable defects.

  • Rounds: 1; findings 0: fixed 0, refuted 0, open 0

  • Findings: none

  • Reviewer: Codex, gpt-6.1-sol, medium effort, shadow

  • Session: 01a1d6d5-0325-72e1-a1d3-f30516210e12, round 1

  • Score: 10/10, The change adds planning documents without changing executable behavior, and no actionable defects were found.

  • Rounds: 1; findings 0: fixed 0, refuted 0, open 0

  • Findings: none

From transcripts on three hosts and five repositories' PR timelines: about
500 answers a day, 74% taken as recommended, and 40% of PR lead time spent
waiting for a merge pick. Plans wider standing picks, picks recorded on the
PR, a landed list, milestone placement by rule, a headless loop with a push,
one cross-repository decisions session, and a quest messages fix.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@kixelated
kixelated marked this pull request as ready for review November 13, 2026 21:35
@coderabbitai

coderabbitai Bot commented Nov 13, 2026

Copy link
Copy Markdown

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 19 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

Run configuration

  • Source mode: Auto
  • Repository: Unavailable
  • Branch: Unavailable
  • File: No configuration file from this repository applied
  • Commit: Unavailable
  • Layers: Organization UI
  • Review profile: CHILL
  • Plan: Advanced
  • Run ID: 3df19437-7311-4e12-bc97-5aa5e6604c38
Commits

Reviewing files that changed from the base of the PR and between cd70476 and 8752084.

Files selected for processing: 9

  • quest/m0/ask-less/README.md
  • quest/m0/ask-less/decisions-session.md
  • quest/m0/ask-less/headless-loop.md
  • quest/m0/ask-less/landed-list.md
  • quest/m0/ask-less/messages-codex-tui.md
  • quest/m0/ask-less/milestone-by-rule.md
  • quest/m0/ask-less/recorded-picks.md
  • quest/m0/ask-less/standing-picks.md
  • quest/m0/mistake-audit.md
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

@pullfrog pullfrog Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

✅ No new issues found.

Reviewed changes — the full diff of the new quest/m0/ask-less/ epic and the mistake-audit.md addition. It changes plans only, and every cross-link target exists.

  • Epic README: goal, recorded decisions, and the measured baseline.
  • Quests: standing-picks, recorded-picks, landed-list, milestone-by-rule, headless-loop, decisions-session, and messages-codex-tui, each with Goal, Plan, and Related or Required sections.
  • mistake-audit.md: a new check counting escapes by standing and maintainer pick, per class.

Pullfrog  | View workflow run | Using claude-sonnet-5-5 | 𝕏

@kixelated

Copy link
Copy Markdown
Collaborator Author

Closing unmerged: the maintainer abandoned the ask-less plan on 2026-11-13 (/quest-run pick: /quest-delete). The messages-codex-tui bug quest it carried will be planned on its own.

(Written by Claude Opus 5.5)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant